For operating modern combustion engines, in particular self-igniting diesel engines, a highly accurate metering of the fuel mass is required at very high injection pressures. The precision of the controlled fuel metering is however limited at modern injection systems on the one hand by the manufacturing accuracy and on the other hand by characteristics of the components of the injection system that change during the lifetime, in particular drifting appearances, which for example cause increased tolerances. In a given case now the aimed accuracy might not be sufficient in order to fulfill the motoric requirements. In particular inaccuracies of the pre-injection can cause significantly increased emissions, for example in partial-load operation, or to conspicuous and therefore disturbing combustion noises depending on the operating point. Therefore there is a great interest to know the injected fuel mass exactly and in particular to know the deviations of the real injection amount that actually occurs during the operation of the combustion engine from the desired injection amount, in order to be able to carry out corresponding corrections of the controlling of the injection valves and thus to be able to ensure a correct injection amount.
A procedure and a device of the category-like type is known from DE 10 2006 061 683.9, at which a correction variable is determined for the injection with the aid of comparison of a measure for the actual amount of the injected fuel of at least one after-injection, which is carried out due to a measure for a default nominal amount of a desired injection, and the measure for the nominal amount of this test-after-injection. The parameters of the controlling of the injection valves are corrected with the aid of the correction variable and thereby a correct injection amount is ensured.
The basis of the procedure is therefore to determine the amount of the after-injections from the difference of the lambda signal in normal operation and the lambda signal in the operation of the additional late and therefore torque-neutral after-injection amount. Now it has proven that the underlying measuring effect is weakened with an increasing size of the combustion engine and that the accuracy and resolution of the lambda probe is too low for a cylinder-individual determination of partial injection amounts in the lower milligram-range.